[linux.git] / fs / exofs / dir.c

/*
 * Copyright (C) 2005, 2006
 * Avishay Traeger ([email protected])
 * Copyright (C) 2008, 2009
 * Boaz Harrosh <[email protected]>
 *
 * Copyrights for code taken from ext2:
 *     Copyright (C) 1992, 1993, 1994, 1995
 *     Remy Card ([email protected])
 *     Laboratoire MASI - Institut Blaise Pascal
 *     Universite Pierre et Marie Curie (Paris VI)
 *     from
 *     linux/fs/minix/inode.c
 *     Copyright (C) 1991, 1992  Linus Torvalds
 *
 * This file is part of exofs.
 *
 * exofs is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation.  Since it is based on ext2, and the only
 * valid version of GPL for the Linux kernel is version 2, the only valid
 * version of GPL for exofs is version 2.
 *
 * exofs is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with exofs; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA  02110-1301  USA
 */

#include <linux/iversion.h>
#include "exofs.h"

static inline unsigned exofs_chunk_size(struct inode *inode)
{
	return inode->i_sb->s_blocksize;
}

static inline void exofs_put_page(struct page *page)
{
	kunmap(page);
	put_page(page);
}

static unsigned exofs_last_byte(struct inode *inode, unsigned long page_nr)
{
	loff_t last_byte = inode->i_size;

	last_byte -= page_nr << PAGE_SHIFT;
	if (last_byte > PAGE_SIZE)
		last_byte = PAGE_SIZE;
	return last_byte;
}

static int exofs_commit_chunk(struct page *page, loff_t pos, unsigned len)
{
	struct address_space *mapping = page->mapping;
	struct inode *dir = mapping->host;
	int err = 0;

	inode_inc_iversion(dir);

	if (!PageUptodate(page))
		SetPageUptodate(page);

	if (pos+len > dir->i_size) {
		i_size_write(dir, pos+len);
		mark_inode_dirty(dir);
	}
	set_page_dirty(page);

	if (IS_DIRSYNC(dir))
		err = write_one_page(page);
	else
		unlock_page(page);

	return err;
}

static bool exofs_check_page(struct page *page)
{
	struct inode *dir = page->mapping->host;
	unsigned chunk_size = exofs_chunk_size(dir);
	char *kaddr = page_address(page);
	unsigned offs, rec_len;
	unsigned limit = PAGE_SIZE;
	struct exofs_dir_entry *p;
	char *error;

	/* if the page is the last one in the directory */
	if ((dir->i_size >> PAGE_SHIFT) == page->index) {
		limit = dir->i_size & ~PAGE_MASK;
		if (limit & (chunk_size - 1))
			goto Ebadsize;
		if (!limit)
			goto out;
	}
	for (offs = 0; offs <= limit - EXOFS_DIR_REC_LEN(1); offs += rec_len) {
		p = (struct exofs_dir_entry *)(kaddr + offs);
		rec_len = le16_to_cpu(p->rec_len);

		if (rec_len < EXOFS_DIR_REC_LEN(1))
			goto Eshort;
		if (rec_len & 3)
			goto Ealign;
		if (rec_len < EXOFS_DIR_REC_LEN(p->name_len))
			goto Enamelen;
		if (((offs + rec_len - 1) ^ offs) & ~(chunk_size-1))
			goto Espan;
	}
	if (offs != limit)
		goto Eend;
out:
	SetPageChecked(page);
	return true;

Ebadsize:
	EXOFS_ERR("ERROR [exofs_check_page]: "
		"size of directory(0x%lx) is not a multiple of chunk size\n",
		dir->i_ino
	);
	goto fail;
Eshort:
	error = "rec_len is smaller than minimal";
	goto bad_entry;
Ealign:
	error = "unaligned directory entry";
	goto bad_entry;
Enamelen:
	error = "rec_len is too small for name_len";
	goto bad_entry;
Espan:
	error = "directory entry across blocks";
	goto bad_entry;
bad_entry:
	EXOFS_ERR(
		"ERROR [exofs_check_page]: bad entry in directory(0x%lx): %s - "
		"offset=%lu, inode=0x%llx, rec_len=%d, name_len=%d\n",
		dir->i_ino, error, (page->index<<PAGE_SHIFT)+offs,
		_LLU(le64_to_cpu(p->inode_no)),
		rec_len, p->name_len);
	goto fail;
Eend:
	p = (struct exofs_dir_entry *)(kaddr + offs);
	EXOFS_ERR("ERROR [exofs_check_page]: "
		"entry in directory(0x%lx) spans the page boundary"
		"offset=%lu, inode=0x%llx\n",
		dir->i_ino, (page->index<<PAGE_SHIFT)+offs,
		_LLU(le64_to_cpu(p->inode_no)));
fail:
	SetPageError(page);
	return false;
}

static struct page *exofs_get_page(struct inode *dir, unsigned long n)
{
	struct address_space *mapping = dir->i_mapping;
	struct page *page = read_mapping_page(mapping, n, NULL);

	if (!IS_ERR(page)) {
		kmap(page);
		if (unlikely(!PageChecked(page))) {
			if (PageError(page) || !exofs_check_page(page))
				goto fail;
		}
	}
	return page;

fail:
	exofs_put_page(page);
	return ERR_PTR(-EIO);
}

static inline int exofs_match(int len, const unsigned char *name,
					struct exofs_dir_entry *de)
{
	if (len != de->name_len)
		return 0;
	if (!de->inode_no)
		return 0;
	return !memcmp(name, de->name, len);
}

static inline
struct exofs_dir_entry *exofs_next_entry(struct exofs_dir_entry *p)
{
	return (struct exofs_dir_entry *)((char *)p + le16_to_cpu(p->rec_len));
}

static inline unsigned
exofs_validate_entry(char *base, unsigned offset, unsigned mask)
{
	struct exofs_dir_entry *de = (struct exofs_dir_entry *)(base + offset);
	struct exofs_dir_entry *p =
			(struct exofs_dir_entry *)(base + (offset&mask));
	while ((char *)p < (char *)de) {
		if (p->rec_len == 0)
			break;
		p = exofs_next_entry(p);
	}
	return (char *)p - base;
}

static unsigned char exofs_filetype_table[EXOFS_FT_MAX] = {
	[EXOFS_FT_UNKNOWN]	= DT_UNKNOWN,
	[EXOFS_FT_REG_FILE]	= DT_REG,
	[EXOFS_FT_DIR]		= DT_DIR,
	[EXOFS_FT_CHRDEV]	= DT_CHR,
	[EXOFS_FT_BLKDEV]	= DT_BLK,
	[EXOFS_FT_FIFO]		= DT_FIFO,
	[EXOFS_FT_SOCK]		= DT_SOCK,
	[EXOFS_FT_SYMLINK]	= DT_LNK,
};

#define S_SHIFT 12
static unsigned char exofs_type_by_mode[S_IFMT >> S_SHIFT] = {
	[S_IFREG >> S_SHIFT]	= EXOFS_FT_REG_FILE,
	[S_IFDIR >> S_SHIFT]	= EXOFS_FT_DIR,
	[S_IFCHR >> S_SHIFT]	= EXOFS_FT_CHRDEV,
	[S_IFBLK >> S_SHIFT]	= EXOFS_FT_BLKDEV,
	[S_IFIFO >> S_SHIFT]	= EXOFS_FT_FIFO,
	[S_IFSOCK >> S_SHIFT]	= EXOFS_FT_SOCK,
	[S_IFLNK >> S_SHIFT]	= EXOFS_FT_SYMLINK,
};

static inline
void exofs_set_de_type(struct exofs_dir_entry *de, struct inode *inode)
{
	umode_t mode = inode->i_mode;
	de->file_type = exofs_type_by_mode[(mode & S_IFMT) >> S_SHIFT];
}

static int
exofs_readdir(struct file *file, struct dir_context *ctx)
{
	loff_t pos = ctx->pos;
	struct inode *inode = file_inode(file);
	unsigned int offset = pos & ~PAGE_MASK;
	unsigned long n = pos >> PAGE_SHIFT;
	unsigned long npages = dir_pages(inode);
	unsigned chunk_mask = ~(exofs_chunk_size(inode)-1);
	bool need_revalidate = !inode_eq_iversion(inode, file->f_version);

	if (pos > inode->i_size - EXOFS_DIR_REC_LEN(1))
		return 0;

	for ( ; n < npages; n++, offset = 0) {
		char *kaddr, *limit;
		struct exofs_dir_entry *de;
		struct page *page = exofs_get_page(inode, n);

		if (IS_ERR(page)) {
			EXOFS_ERR("ERROR: bad page in directory(0x%lx)\n",
				  inode->i_ino);
			ctx->pos += PAGE_SIZE - offset;
			return PTR_ERR(page);
		}
		kaddr = page_address(page);
		if (unlikely(need_revalidate)) {
			if (offset) {
				offset = exofs_validate_entry(kaddr, offset,
								chunk_mask);
				ctx->pos = (n<<PAGE_SHIFT) + offset;
			}
			file->f_version = inode_query_iversion(inode);
			need_revalidate = false;
		}
		de = (struct exofs_dir_entry *)(kaddr + offset);
		limit = kaddr + exofs_last_byte(inode, n) -
							EXOFS_DIR_REC_LEN(1);
		for (; (char *)de <= limit; de = exofs_next_entry(de)) {
			if (de->rec_len == 0) {
				EXOFS_ERR("ERROR: "
				     "zero-length entry in directory(0x%lx)\n",
				     inode->i_ino);
				exofs_put_page(page);
				return -EIO;
			}
			if (de->inode_no) {
				unsigned char t;

				if (de->file_type < EXOFS_FT_MAX)
					t = exofs_filetype_table[de->file_type];
				else
					t = DT_UNKNOWN;

				if (!dir_emit(ctx, de->name, de->name_len,
						le64_to_cpu(de->inode_no),
						t)) {
					exofs_put_page(page);
					return 0;
				}
			}
			ctx->pos += le16_to_cpu(de->rec_len);
		}
		exofs_put_page(page);
	}
	return 0;
}

struct exofs_dir_entry *exofs_find_entry(struct inode *dir,
			struct dentry *dentry, struct page **res_page)
{
	const unsigned char *name = dentry->d_name.name;
	int namelen = dentry->d_name.len;
	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	unsigned long start, n;
	unsigned long npages = dir_pages(dir);
	struct page *page = NULL;
	struct exofs_i_info *oi = exofs_i(dir);
	struct exofs_dir_entry *de;

	if (npages == 0)
		goto out;

	*res_page = NULL;

	start = oi->i_dir_start_lookup;
	if (start >= npages)
		start = 0;
	n = start;
	do {
		char *kaddr;
		page = exofs_get_page(dir, n);
		if (!IS_ERR(page)) {
			kaddr = page_address(page);
			de = (struct exofs_dir_entry *) kaddr;
			kaddr += exofs_last_byte(dir, n) - reclen;
			while ((char *) de <= kaddr) {
				if (de->rec_len == 0) {
					EXOFS_ERR("ERROR: zero-length entry in "
						  "directory(0x%lx)\n",
						  dir->i_ino);
					exofs_put_page(page);
					goto out;
				}
				if (exofs_match(namelen, name, de))
					goto found;
				de = exofs_next_entry(de);
			}
			exofs_put_page(page);
		}
		if (++n >= npages)
			n = 0;
	} while (n != start);
out:
	return NULL;

found:
	*res_page = page;
	oi->i_dir_start_lookup = n;
	return de;
}

struct exofs_dir_entry *exofs_dotdot(struct inode *dir, struct page **p)
{
	struct page *page = exofs_get_page(dir, 0);
	struct exofs_dir_entry *de = NULL;

	if (!IS_ERR(page)) {
		de = exofs_next_entry(
				(struct exofs_dir_entry *)page_address(page));
		*p = page;
	}
	return de;
}

ino_t exofs_parent_ino(struct dentry *child)
{
	struct page *page;
	struct exofs_dir_entry *de;
	ino_t ino;

	de = exofs_dotdot(d_inode(child), &page);
	if (!de)
		return 0;

	ino = le64_to_cpu(de->inode_no);
	exofs_put_page(page);
	return ino;
}

ino_t exofs_inode_by_name(struct inode *dir, struct dentry *dentry)
{
	ino_t res = 0;
	struct exofs_dir_entry *de;
	struct page *page;

	de = exofs_find_entry(dir, dentry, &page);
	if (de) {
		res = le64_to_cpu(de->inode_no);
		exofs_put_page(page);
	}
	return res;
}

int exofs_set_link(struct inode *dir, struct exofs_dir_entry *de,
			struct page *page, struct inode *inode)
{
	loff_t pos = page_offset(page) +
			(char *) de - (char *) page_address(page);
	unsigned len = le16_to_cpu(de->rec_len);
	int err;

	lock_page(page);
	err = exofs_write_begin(NULL, page->mapping, pos, len, 0, &page, NULL);
	if (err)
		EXOFS_ERR("exofs_set_link: exofs_write_begin FAILED => %d\n",
			  err);

	de->inode_no = cpu_to_le64(inode->i_ino);
	exofs_set_de_type(de, inode);
	if (likely(!err))
		err = exofs_commit_chunk(page, pos, len);
	exofs_put_page(page);
	dir->i_mtime = dir->i_ctime = current_time(dir);
	mark_inode_dirty(dir);
	return err;
}

int exofs_add_link(struct dentry *dentry, struct inode *inode)
{
	struct inode *dir = d_inode(dentry->d_parent);
	const unsigned char *name = dentry->d_name.name;
	int namelen = dentry->d_name.len;
	unsigned chunk_size = exofs_chunk_size(dir);
	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	unsigned short rec_len, name_len;
	struct page *page = NULL;
	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	struct exofs_dir_entry *de;
	unsigned long npages = dir_pages(dir);
	unsigned long n;
	char *kaddr;
	loff_t pos;
	int err;

	for (n = 0; n <= npages; n++) {
		char *dir_end;

		page = exofs_get_page(dir, n);
		err = PTR_ERR(page);
		if (IS_ERR(page))
			goto out;
		lock_page(page);
		kaddr = page_address(page);
		dir_end = kaddr + exofs_last_byte(dir, n);
		de = (struct exofs_dir_entry *)kaddr;
		kaddr += PAGE_SIZE - reclen;
		while ((char *)de <= kaddr) {
			if ((char *)de == dir_end) {
				name_len = 0;
				rec_len = chunk_size;
				de->rec_len = cpu_to_le16(chunk_size);
				de->inode_no = 0;
				goto got_it;
			}
			if (de->rec_len == 0) {
				EXOFS_ERR("ERROR: exofs_add_link: "
				      "zero-length entry in directory(0x%lx)\n",
				      inode->i_ino);
				err = -EIO;
				goto out_unlock;
			}
			err = -EEXIST;
			if (exofs_match(namelen, name, de))
				goto out_unlock;
			name_len = EXOFS_DIR_REC_LEN(de->name_len);
			rec_len = le16_to_cpu(de->rec_len);
			if (!de->inode_no && rec_len >= reclen)
				goto got_it;
			if (rec_len >= name_len + reclen)
				goto got_it;
			de = (struct exofs_dir_entry *) ((char *) de + rec_len);
		}
		unlock_page(page);
		exofs_put_page(page);
	}

	EXOFS_ERR("exofs_add_link: BAD dentry=%p or inode=0x%lx\n",
		  dentry, inode->i_ino);
	return -EINVAL;

got_it:
	pos = page_offset(page) +
		(char *)de - (char *)page_address(page);
	err = exofs_write_begin(NULL, page->mapping, pos, rec_len, 0,
							&page, NULL);
	if (err)
		goto out_unlock;
	if (de->inode_no) {
		struct exofs_dir_entry *de1 =
			(struct exofs_dir_entry *)((char *)de + name_len);
		de1->rec_len = cpu_to_le16(rec_len - name_len);
		de->rec_len = cpu_to_le16(name_len);
		de = de1;
	}
	de->name_len = namelen;
	memcpy(de->name, name, namelen);
	de->inode_no = cpu_to_le64(inode->i_ino);
	exofs_set_de_type(de, inode);
	err = exofs_commit_chunk(page, pos, rec_len);
	dir->i_mtime = dir->i_ctime = current_time(dir);
	mark_inode_dirty(dir);
	sbi->s_numfiles++;

out_put:
	exofs_put_page(page);
out:
	return err;
out_unlock:
	unlock_page(page);
	goto out_put;
}

int exofs_delete_entry(struct exofs_dir_entry *dir, struct page *page)
{
	struct address_space *mapping = page->mapping;
	struct inode *inode = mapping->host;
	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	char *kaddr = page_address(page);
	unsigned from = ((char *)dir - kaddr) & ~(exofs_chunk_size(inode)-1);
	unsigned to = ((char *)dir - kaddr) + le16_to_cpu(dir->rec_len);
	loff_t pos;
	struct exofs_dir_entry *pde = NULL;
	struct exofs_dir_entry *de = (struct exofs_dir_entry *) (kaddr + from);
	int err;

	while (de < dir) {
		if (de->rec_len == 0) {
			EXOFS_ERR("ERROR: exofs_delete_entry:"
				  "zero-length entry in directory(0x%lx)\n",
				  inode->i_ino);
			err = -EIO;
			goto out;
		}
		pde = de;
		de = exofs_next_entry(de);
	}
	if (pde)
		from = (char *)pde - (char *)page_address(page);
	pos = page_offset(page) + from;
	lock_page(page);
	err = exofs_write_begin(NULL, page->mapping, pos, to - from, 0,
							&page, NULL);
	if (err)
		EXOFS_ERR("exofs_delete_entry: exofs_write_begin FAILED => %d\n",
			  err);
	if (pde)
		pde->rec_len = cpu_to_le16(to - from);
	dir->inode_no = 0;
	if (likely(!err))
		err = exofs_commit_chunk(page, pos, to - from);
	inode->i_ctime = inode->i_mtime = current_time(inode);
	mark_inode_dirty(inode);
	sbi->s_numfiles--;
out:
	exofs_put_page(page);
	return err;
}

/* kept aligned on 4 bytes */
#define THIS_DIR ".\0\0"
#define PARENT_DIR "..\0"

int exofs_make_empty(struct inode *inode, struct inode *parent)
{
	struct address_space *mapping = inode->i_mapping;
	struct page *page = grab_cache_page(mapping, 0);
	unsigned chunk_size = exofs_chunk_size(inode);
	struct exofs_dir_entry *de;
	int err;
	void *kaddr;

	if (!page)
		return -ENOMEM;

	err = exofs_write_begin(NULL, page->mapping, 0, chunk_size, 0,
							&page, NULL);
	if (err) {
		unlock_page(page);
		goto fail;
	}

	kaddr = kmap_atomic(page);
	de = (struct exofs_dir_entry *)kaddr;
	de->name_len = 1;
	de->rec_len = cpu_to_le16(EXOFS_DIR_REC_LEN(1));
	memcpy(de->name, THIS_DIR, sizeof(THIS_DIR));
	de->inode_no = cpu_to_le64(inode->i_ino);
	exofs_set_de_type(de, inode);

	de = (struct exofs_dir_entry *)(kaddr + EXOFS_DIR_REC_LEN(1));
	de->name_len = 2;
	de->rec_len = cpu_to_le16(chunk_size - EXOFS_DIR_REC_LEN(1));
	de->inode_no = cpu_to_le64(parent->i_ino);
	memcpy(de->name, PARENT_DIR, sizeof(PARENT_DIR));
	exofs_set_de_type(de, inode);
	kunmap_atomic(kaddr);
	err = exofs_commit_chunk(page, 0, chunk_size);
fail:
	put_page(page);
	return err;
}

int exofs_empty_dir(struct inode *inode)
{
	struct page *page = NULL;
	unsigned long i, npages = dir_pages(inode);

	for (i = 0; i < npages; i++) {
		char *kaddr;
		struct exofs_dir_entry *de;
		page = exofs_get_page(inode, i);

		if (IS_ERR(page))
			continue;

		kaddr = page_address(page);
		de = (struct exofs_dir_entry *)kaddr;
		kaddr += exofs_last_byte(inode, i) - EXOFS_DIR_REC_LEN(1);

		while ((char *)de <= kaddr) {
			if (de->rec_len == 0) {
				EXOFS_ERR("ERROR: exofs_empty_dir: "
					  "zero-length directory entry"
					  "kaddr=%p, de=%p\n", kaddr, de);
				goto not_empty;
			}
			if (de->inode_no != 0) {
				/* check for . and .. */
				if (de->name[0] != '.')
					goto not_empty;
				if (de->name_len > 2)
					goto not_empty;
				if (de->name_len < 2) {
					if (le64_to_cpu(de->inode_no) !=
					    inode->i_ino)
						goto not_empty;
				} else if (de->name[1] != '.')
					goto not_empty;
			}
			de = exofs_next_entry(de);
		}
		exofs_put_page(page);
	}
	return 1;

not_empty:
	exofs_put_page(page);
	return 0;
}

const struct file_operations exofs_dir_operations = {
	.llseek		= generic_file_llseek,
	.read		= generic_read_dir,
	.iterate_shared	= exofs_readdir,
};
Commit	Line	Data
e6af00f1 BH	1	/*
e6af00f1 BH	2	* Copyright (C) 2005, 2006
27d2e149	3	* Avishay Traeger ([email protected])
e6af00f1	4	* Copyright (C) 2008, 2009
aa281ac6	5	* Boaz Harrosh <[email protected]>
e6af00f1 BH	6	*
	7	* Copyrights for code taken from ext2:
	8	* Copyright (C) 1992, 1993, 1994, 1995
	9	* Remy Card ([email protected])
	10	* Laboratoire MASI - Institut Blaise Pascal
	11	* Universite Pierre et Marie Curie (Paris VI)
	12	* from
	13	* linux/fs/minix/inode.c
	14	* Copyright (C) 1991, 1992 Linus Torvalds
	15	*
	16	* This file is part of exofs.
	17	*
	18	* exofs is free software; you can redistribute it and/or modify
	19	* it under the terms of the GNU General Public License as published by
	20	* the Free Software Foundation. Since it is based on ext2, and the only
	21	* valid version of GPL for the Linux kernel is version 2, the only valid
	22	* version of GPL for exofs is version 2.
	23	*
	24	* exofs is distributed in the hope that it will be useful,
	25	* but WITHOUT ANY WARRANTY; without even the implied warranty of
	26	* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
	27	* GNU General Public License for more details.
	28	*
	29	* You should have received a copy of the GNU General Public License
	30	* along with exofs; if not, write to the Free Software
	31	* Foundation, Inc., 51 Franklin St, Fifth Floor, Boston, MA 02110-1301 USA
	32	*/
	33
317bc947	34	#include <linux/iversion.h>
e6af00f1 BH	35	#include "exofs.h"
	36
	37	static inline unsigned exofs_chunk_size(struct inode *inode)
	38	{
	39	return inode->i_sb->s_blocksize;
	40	}
	41
	42	static inline void exofs_put_page(struct page *page)
	43	{
	44	kunmap(page);
09cbfeaf	45	put_page(page);
e6af00f1 BH	46	}
e6af00f1 BH	47
e6af00f1 BH	48	static unsigned exofs_last_byte(struct inode *inode, unsigned long page_nr)
	49	{
	50	loff_t last_byte = inode->i_size;
	51
09cbfeaf KS	52	last_byte -= page_nr << PAGE_SHIFT;
	53	if (last_byte > PAGE_SIZE)
	54	last_byte = PAGE_SIZE;
e6af00f1 BH	55	return last_byte;
	56	}
	57
	58	static int exofs_commit_chunk(struct page *page, loff_t pos, unsigned len)
	59	{
	60	struct address_space *mapping = page->mapping;
	61	struct inode *dir = mapping->host;
	62	int err = 0;
	63
317bc947	64	inode_inc_iversion(dir);
e6af00f1 BH	65
	66	if (!PageUptodate(page))
	67	SetPageUptodate(page);
	68
	69	if (pos+len > dir->i_size) {
	70	i_size_write(dir, pos+len);
	71	mark_inode_dirty(dir);
	72	}
	73	set_page_dirty(page);
	74
	75	if (IS_DIRSYNC(dir))
2b69c828	76	err = write_one_page(page);
e6af00f1 BH	77	else
	78	unlock_page(page);
	79
	80	return err;
	81	}
	82
be5b82db	83	static bool exofs_check_page(struct page *page)
e6af00f1 BH	84	{
	85	struct inode *dir = page->mapping->host;
	86	unsigned chunk_size = exofs_chunk_size(dir);
	87	char *kaddr = page_address(page);
	88	unsigned offs, rec_len;
09cbfeaf	89	unsigned limit = PAGE_SIZE;
e6af00f1 BH	90	struct exofs_dir_entry *p;
	91	char *error;
	92
	93	/* if the page is the last one in the directory */
09cbfeaf KS	94	if ((dir->i_size >> PAGE_SHIFT) == page->index) {
09cbfeaf KS	95	limit = dir->i_size & ~PAGE_MASK;
e6af00f1 BH	96	if (limit & (chunk_size - 1))
	97	goto Ebadsize;
	98	if (!limit)
	99	goto out;
	100	}
	101	for (offs = 0; offs <= limit - EXOFS_DIR_REC_LEN(1); offs += rec_len) {
	102	p = (struct exofs_dir_entry *)(kaddr + offs);
	103	rec_len = le16_to_cpu(p->rec_len);
	104
	105	if (rec_len < EXOFS_DIR_REC_LEN(1))
	106	goto Eshort;
	107	if (rec_len & 3)
	108	goto Ealign;
	109	if (rec_len < EXOFS_DIR_REC_LEN(p->name_len))
	110	goto Enamelen;
	111	if (((offs + rec_len - 1) ^ offs) & ~(chunk_size-1))
	112	goto Espan;
	113	}
	114	if (offs != limit)
	115	goto Eend;
	116	out:
	117	SetPageChecked(page);
be5b82db	118	return true;
e6af00f1 BH	119
	120	Ebadsize:
	121	EXOFS_ERR("ERROR [exofs_check_page]: "
0a935519	122	"size of directory(0x%lx) is not a multiple of chunk size\n",
e6af00f1 BH	123	dir->i_ino
	124	);
	125	goto fail;
	126	Eshort:
	127	error = "rec_len is smaller than minimal";
	128	goto bad_entry;
	129	Ealign:
	130	error = "unaligned directory entry";
	131	goto bad_entry;
	132	Enamelen:
	133	error = "rec_len is too small for name_len";
	134	goto bad_entry;
	135	Espan:
	136	error = "directory entry across blocks";
	137	goto bad_entry;
	138	bad_entry:
	139	EXOFS_ERR(
0a935519	140	"ERROR [exofs_check_page]: bad entry in directory(0x%lx): %s - "
14f947c8	141	"offset=%lu, inode=0x%llx, rec_len=%d, name_len=%d\n",
09cbfeaf	142	dir->i_ino, error, (page->index<<PAGE_SHIFT)+offs,
e6af00f1 BH	143	_LLU(le64_to_cpu(p->inode_no)),
	144	rec_len, p->name_len);
	145	goto fail;
	146	Eend:
	147	p = (struct exofs_dir_entry *)(kaddr + offs);
	148	EXOFS_ERR("ERROR [exofs_check_page]: "
0a935519 BH	149	"entry in directory(0x%lx) spans the page boundary"
0a935519 BH	150	"offset=%lu, inode=0x%llx\n",
09cbfeaf	151	dir->i_ino, (page->index<<PAGE_SHIFT)+offs,
e6af00f1 BH	152	_LLU(le64_to_cpu(p->inode_no)));
e6af00f1 BH	153	fail:
e6af00f1	154	SetPageError(page);
be5b82db	155	return false;
e6af00f1 BH	156	}
	157
	158	static struct page exofs_get_page(struct inode dir, unsigned long n)
	159	{
	160	struct address_space *mapping = dir->i_mapping;
	161	struct page *page = read_mapping_page(mapping, n, NULL);
	162
	163	if (!IS_ERR(page)) {
	164	kmap(page);
be5b82db AV	165	if (unlikely(!PageChecked(page))) {
	166	if (PageError(page) \|\| !exofs_check_page(page))
	167	goto fail;
	168	}
e6af00f1 BH	169	}
	170	return page;
	171
	172	fail:
	173	exofs_put_page(page);
	174	return ERR_PTR(-EIO);
	175	}
	176
	177	static inline int exofs_match(int len, const unsigned char *name,
	178	struct exofs_dir_entry *de)
	179	{
	180	if (len != de->name_len)
	181	return 0;
	182	if (!de->inode_no)
	183	return 0;
	184	return !memcmp(name, de->name, len);
	185	}
	186
	187	static inline
	188	struct exofs_dir_entry exofs_next_entry(struct exofs_dir_entry p)
	189	{
	190	return (struct exofs_dir_entry )((char )p + le16_to_cpu(p->rec_len));
	191	}
	192
	193	static inline unsigned
	194	exofs_validate_entry(char *base, unsigned offset, unsigned mask)
	195	{
	196	struct exofs_dir_entry de = (struct exofs_dir_entry )(base + offset);
	197	struct exofs_dir_entry *p =
	198	(struct exofs_dir_entry *)(base + (offset&mask));
	199	while ((char )p < (char )de) {
	200	if (p->rec_len == 0)
	201	break;
	202	p = exofs_next_entry(p);
	203	}
	204	return (char *)p - base;
	205	}
	206
	207	static unsigned char exofs_filetype_table[EXOFS_FT_MAX] = {
	208	[EXOFS_FT_UNKNOWN] = DT_UNKNOWN,
	209	[EXOFS_FT_REG_FILE] = DT_REG,
	210	[EXOFS_FT_DIR] = DT_DIR,
	211	[EXOFS_FT_CHRDEV] = DT_CHR,
	212	[EXOFS_FT_BLKDEV] = DT_BLK,
	213	[EXOFS_FT_FIFO] = DT_FIFO,
	214	[EXOFS_FT_SOCK] = DT_SOCK,
	215	[EXOFS_FT_SYMLINK] = DT_LNK,
	216	};
	217
	218	#define S_SHIFT 12
	219	static unsigned char exofs_type_by_mode[S_IFMT >> S_SHIFT] = {
	220	[S_IFREG >> S_SHIFT] = EXOFS_FT_REG_FILE,
	221	[S_IFDIR >> S_SHIFT] = EXOFS_FT_DIR,
	222	[S_IFCHR >> S_SHIFT] = EXOFS_FT_CHRDEV,
	223	[S_IFBLK >> S_SHIFT] = EXOFS_FT_BLKDEV,
	224	[S_IFIFO >> S_SHIFT] = EXOFS_FT_FIFO,
	225	[S_IFSOCK >> S_SHIFT] = EXOFS_FT_SOCK,
	226	[S_IFLNK >> S_SHIFT] = EXOFS_FT_SYMLINK,
	227	};
	228
	229	static inline
	230	void exofs_set_de_type(struct exofs_dir_entry de, struct inode inode)
	231	{
bef41c26	232	umode_t mode = inode->i_mode;
e6af00f1 BH	233	de->file_type = exofs_type_by_mode[(mode & S_IFMT) >> S_SHIFT];
	234	}
	235
	236	static int
75811d4f	237	exofs_readdir(struct file file, struct dir_context ctx)
e6af00f1	238	{
75811d4f AV	239	loff_t pos = ctx->pos;
75811d4f AV	240	struct inode *inode = file_inode(file);
09cbfeaf KS	241	unsigned int offset = pos & ~PAGE_MASK;
09cbfeaf KS	242	unsigned long n = pos >> PAGE_SHIFT;
e6af00f1 BH	243	unsigned long npages = dir_pages(inode);
e6af00f1 BH	244	unsigned chunk_mask = ~(exofs_chunk_size(inode)-1);
c472c07b	245	bool need_revalidate = !inode_eq_iversion(inode, file->f_version);
e6af00f1 BH	246
	247	if (pos > inode->i_size - EXOFS_DIR_REC_LEN(1))
	248	return 0;
	249
e6af00f1 BH	250	for ( ; n < npages; n++, offset = 0) {
	251	char kaddr, limit;
	252	struct exofs_dir_entry *de;
	253	struct page *page = exofs_get_page(inode, n);
	254
	255	if (IS_ERR(page)) {
0a935519 BH	256	EXOFS_ERR("ERROR: bad page in directory(0x%lx)\n",
0a935519 BH	257	inode->i_ino);
09cbfeaf	258	ctx->pos += PAGE_SIZE - offset;
e6af00f1 BH	259	return PTR_ERR(page);
	260	}
	261	kaddr = page_address(page);
	262	if (unlikely(need_revalidate)) {
	263	if (offset) {
	264	offset = exofs_validate_entry(kaddr, offset,
	265	chunk_mask);
09cbfeaf	266	ctx->pos = (n<<PAGE_SHIFT) + offset;
e6af00f1	267	}
317bc947 JL	268	file->f_version = inode_query_iversion(inode);
317bc947 JL	269	need_revalidate = false;
e6af00f1 BH	270	}
	271	de = (struct exofs_dir_entry *)(kaddr + offset);
	272	limit = kaddr + exofs_last_byte(inode, n) -
	273	EXOFS_DIR_REC_LEN(1);
	274	for (; (char *)de <= limit; de = exofs_next_entry(de)) {
	275	if (de->rec_len == 0) {
	276	EXOFS_ERR("ERROR: "
0a935519 BH	277	"zero-length entry in directory(0x%lx)\n",
0a935519 BH	278	inode->i_ino);
e6af00f1 BH	279	exofs_put_page(page);
	280	return -EIO;
	281	}
	282	if (de->inode_no) {
75811d4f	283	unsigned char t;
e6af00f1	284
75811d4f AV	285	if (de->file_type < EXOFS_FT_MAX)
	286	t = exofs_filetype_table[de->file_type];
	287	else
	288	t = DT_UNKNOWN;
e6af00f1	289
75811d4f	290	if (!dir_emit(ctx, de->name, de->name_len,
e6af00f1	291	le64_to_cpu(de->inode_no),
75811d4f	292	t)) {
e6af00f1 BH	293	exofs_put_page(page);
	294	return 0;
	295	}
	296	}
75811d4f	297	ctx->pos += le16_to_cpu(de->rec_len);
e6af00f1 BH	298	}
	299	exofs_put_page(page);
	300	}
e6af00f1 BH	301	return 0;
	302	}
	303
	304	struct exofs_dir_entry exofs_find_entry(struct inode dir,
	305	struct dentry dentry, struct page *res_page)
	306	{
	307	const unsigned char *name = dentry->d_name.name;
	308	int namelen = dentry->d_name.len;
	309	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	310	unsigned long start, n;
	311	unsigned long npages = dir_pages(dir);
	312	struct page *page = NULL;
	313	struct exofs_i_info *oi = exofs_i(dir);
	314	struct exofs_dir_entry *de;
	315
	316	if (npages == 0)
	317	goto out;
	318
	319	*res_page = NULL;
	320
	321	start = oi->i_dir_start_lookup;
	322	if (start >= npages)
	323	start = 0;
	324	n = start;
	325	do {
	326	char *kaddr;
	327	page = exofs_get_page(dir, n);
	328	if (!IS_ERR(page)) {
	329	kaddr = page_address(page);
	330	de = (struct exofs_dir_entry *) kaddr;
	331	kaddr += exofs_last_byte(dir, n) - reclen;
	332	while ((char *) de <= kaddr) {
	333	if (de->rec_len == 0) {
0a935519 BH	334	EXOFS_ERR("ERROR: zero-length entry in "
	335	"directory(0x%lx)\n",
	336	dir->i_ino);
e6af00f1 BH	337	exofs_put_page(page);
	338	goto out;
	339	}
	340	if (exofs_match(namelen, name, de))
	341	goto found;
	342	de = exofs_next_entry(de);
	343	}
	344	exofs_put_page(page);
	345	}
	346	if (++n >= npages)
	347	n = 0;
	348	} while (n != start);
	349	out:
	350	return NULL;
	351
	352	found:
	353	*res_page = page;
	354	oi->i_dir_start_lookup = n;
	355	return de;
	356	}
	357
	358	struct exofs_dir_entry exofs_dotdot(struct inode dir, struct page **p)
	359	{
	360	struct page *page = exofs_get_page(dir, 0);
	361	struct exofs_dir_entry *de = NULL;
	362
	363	if (!IS_ERR(page)) {
	364	de = exofs_next_entry(
	365	(struct exofs_dir_entry *)page_address(page));
	366	*p = page;
	367	}
	368	return de;
	369	}
	370
8cf74b39 BH	371	ino_t exofs_parent_ino(struct dentry *child)
	372	{
	373	struct page *page;
	374	struct exofs_dir_entry *de;
	375	ino_t ino;
	376
2b0143b5	377	de = exofs_dotdot(d_inode(child), &page);
8cf74b39 BH	378	if (!de)
	379	return 0;
	380
	381	ino = le64_to_cpu(de->inode_no);
	382	exofs_put_page(page);
	383	return ino;
	384	}
	385
e6af00f1 BH	386	ino_t exofs_inode_by_name(struct inode dir, struct dentry dentry)
	387	{
	388	ino_t res = 0;
	389	struct exofs_dir_entry *de;
	390	struct page *page;
	391
	392	de = exofs_find_entry(dir, dentry, &page);
	393	if (de) {
	394	res = le64_to_cpu(de->inode_no);
	395	exofs_put_page(page);
	396	}
	397	return res;
	398	}
	399
	400	int exofs_set_link(struct inode dir, struct exofs_dir_entry de,
	401	struct page page, struct inode inode)
	402	{
	403	loff_t pos = page_offset(page) +
	404	(char ) de - (char ) page_address(page);
	405	unsigned len = le16_to_cpu(de->rec_len);
	406	int err;
	407
	408	lock_page(page);
c718a975	409	err = exofs_write_begin(NULL, page->mapping, pos, len, 0, &page, NULL);
e6af00f1	410	if (err)
571f7f46	411	EXOFS_ERR("exofs_set_link: exofs_write_begin FAILED => %d\n",
e6af00f1 BH	412	err);
	413
	414	de->inode_no = cpu_to_le64(inode->i_ino);
	415	exofs_set_de_type(de, inode);
	416	if (likely(!err))
	417	err = exofs_commit_chunk(page, pos, len);
	418	exofs_put_page(page);
078cd827	419	dir->i_mtime = dir->i_ctime = current_time(dir);
e6af00f1 BH	420	mark_inode_dirty(dir);
	421	return err;
	422	}
	423
	424	int exofs_add_link(struct dentry dentry, struct inode inode)
	425	{
2b0143b5	426	struct inode *dir = d_inode(dentry->d_parent);
e6af00f1 BH	427	const unsigned char *name = dentry->d_name.name;
	428	int namelen = dentry->d_name.len;
	429	unsigned chunk_size = exofs_chunk_size(dir);
	430	unsigned reclen = EXOFS_DIR_REC_LEN(namelen);
	431	unsigned short rec_len, name_len;
	432	struct page *page = NULL;
	433	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	434	struct exofs_dir_entry *de;
	435	unsigned long npages = dir_pages(dir);
	436	unsigned long n;
	437	char *kaddr;
	438	loff_t pos;
	439	int err;
	440
	441	for (n = 0; n <= npages; n++) {
	442	char *dir_end;
	443
	444	page = exofs_get_page(dir, n);
	445	err = PTR_ERR(page);
	446	if (IS_ERR(page))
	447	goto out;
	448	lock_page(page);
	449	kaddr = page_address(page);
	450	dir_end = kaddr + exofs_last_byte(dir, n);
	451	de = (struct exofs_dir_entry *)kaddr;
09cbfeaf	452	kaddr += PAGE_SIZE - reclen;
e6af00f1 BH	453	while ((char *)de <= kaddr) {
	454	if ((char *)de == dir_end) {
	455	name_len = 0;
	456	rec_len = chunk_size;
	457	de->rec_len = cpu_to_le16(chunk_size);
	458	de->inode_no = 0;
	459	goto got_it;
	460	}
	461	if (de->rec_len == 0) {
	462	EXOFS_ERR("ERROR: exofs_add_link: "
0a935519 BH	463	"zero-length entry in directory(0x%lx)\n",
0a935519 BH	464	inode->i_ino);
e6af00f1 BH	465	err = -EIO;
	466	goto out_unlock;
	467	}
	468	err = -EEXIST;
	469	if (exofs_match(namelen, name, de))
	470	goto out_unlock;
	471	name_len = EXOFS_DIR_REC_LEN(de->name_len);
	472	rec_len = le16_to_cpu(de->rec_len);
	473	if (!de->inode_no && rec_len >= reclen)
	474	goto got_it;
	475	if (rec_len >= name_len + reclen)
	476	goto got_it;
	477	de = (struct exofs_dir_entry ) ((char ) de + rec_len);
	478	}
	479	unlock_page(page);
	480	exofs_put_page(page);
	481	}
	482
0a935519 BH	483	EXOFS_ERR("exofs_add_link: BAD dentry=%p or inode=0x%lx\n",
0a935519 BH	484	dentry, inode->i_ino);
e6af00f1 BH	485	return -EINVAL;
	486
	487	got_it:
	488	pos = page_offset(page) +
	489	(char )de - (char )page_address(page);
	490	err = exofs_write_begin(NULL, page->mapping, pos, rec_len, 0,
	491	&page, NULL);
	492	if (err)
	493	goto out_unlock;
	494	if (de->inode_no) {
	495	struct exofs_dir_entry *de1 =
	496	(struct exofs_dir_entry )((char )de + name_len);
	497	de1->rec_len = cpu_to_le16(rec_len - name_len);
	498	de->rec_len = cpu_to_le16(name_len);
	499	de = de1;
	500	}
	501	de->name_len = namelen;
	502	memcpy(de->name, name, namelen);
	503	de->inode_no = cpu_to_le64(inode->i_ino);
	504	exofs_set_de_type(de, inode);
	505	err = exofs_commit_chunk(page, pos, rec_len);
078cd827	506	dir->i_mtime = dir->i_ctime = current_time(dir);
e6af00f1 BH	507	mark_inode_dirty(dir);
	508	sbi->s_numfiles++;
	509
	510	out_put:
	511	exofs_put_page(page);
	512	out:
	513	return err;
	514	out_unlock:
	515	unlock_page(page);
	516	goto out_put;
	517	}
	518
	519	int exofs_delete_entry(struct exofs_dir_entry dir, struct page page)
	520	{
	521	struct address_space *mapping = page->mapping;
	522	struct inode *inode = mapping->host;
	523	struct exofs_sb_info *sbi = inode->i_sb->s_fs_info;
	524	char *kaddr = page_address(page);
	525	unsigned from = ((char *)dir - kaddr) & ~(exofs_chunk_size(inode)-1);
	526	unsigned to = ((char *)dir - kaddr) + le16_to_cpu(dir->rec_len);
	527	loff_t pos;
	528	struct exofs_dir_entry *pde = NULL;
	529	struct exofs_dir_entry de = (struct exofs_dir_entry ) (kaddr + from);
	530	int err;
	531
	532	while (de < dir) {
	533	if (de->rec_len == 0) {
	534	EXOFS_ERR("ERROR: exofs_delete_entry:"
0a935519 BH	535	"zero-length entry in directory(0x%lx)\n",
0a935519 BH	536	inode->i_ino);
e6af00f1 BH	537	err = -EIO;
	538	goto out;
	539	}
	540	pde = de;
	541	de = exofs_next_entry(de);
	542	}
	543	if (pde)
	544	from = (char )pde - (char )page_address(page);
	545	pos = page_offset(page) + from;
	546	lock_page(page);
	547	err = exofs_write_begin(NULL, page->mapping, pos, to - from, 0,
	548	&page, NULL);
	549	if (err)
571f7f46	550	EXOFS_ERR("exofs_delete_entry: exofs_write_begin FAILED => %d\n",
e6af00f1 BH	551	err);
	552	if (pde)
	553	pde->rec_len = cpu_to_le16(to - from);
	554	dir->inode_no = 0;
	555	if (likely(!err))
	556	err = exofs_commit_chunk(page, pos, to - from);
078cd827	557	inode->i_ctime = inode->i_mtime = current_time(inode);
e6af00f1 BH	558	mark_inode_dirty(inode);
	559	sbi->s_numfiles--;
	560	out:
	561	exofs_put_page(page);
	562	return err;
	563	}
	564
	565	/* kept aligned on 4 bytes */
	566	#define THIS_DIR ".\0\0"
	567	#define PARENT_DIR "..\0"
	568
	569	int exofs_make_empty(struct inode inode, struct inode parent)
	570	{
	571	struct address_space *mapping = inode->i_mapping;
	572	struct page *page = grab_cache_page(mapping, 0);
	573	unsigned chunk_size = exofs_chunk_size(inode);
	574	struct exofs_dir_entry *de;
	575	int err;
	576	void *kaddr;
	577
	578	if (!page)
	579	return -ENOMEM;
	580
	581	err = exofs_write_begin(NULL, page->mapping, 0, chunk_size, 0,
	582	&page, NULL);
	583	if (err) {
	584	unlock_page(page);
	585	goto fail;
	586	}
	587
bf7014b6	588	kaddr = kmap_atomic(page);
e6af00f1 BH	589	de = (struct exofs_dir_entry *)kaddr;
	590	de->name_len = 1;
	591	de->rec_len = cpu_to_le16(EXOFS_DIR_REC_LEN(1));
	592	memcpy(de->name, THIS_DIR, sizeof(THIS_DIR));
	593	de->inode_no = cpu_to_le64(inode->i_ino);
	594	exofs_set_de_type(de, inode);
	595
	596	de = (struct exofs_dir_entry *)(kaddr + EXOFS_DIR_REC_LEN(1));
	597	de->name_len = 2;
	598	de->rec_len = cpu_to_le16(chunk_size - EXOFS_DIR_REC_LEN(1));
	599	de->inode_no = cpu_to_le64(parent->i_ino);
	600	memcpy(de->name, PARENT_DIR, sizeof(PARENT_DIR));
	601	exofs_set_de_type(de, inode);
bf7014b6	602	kunmap_atomic(kaddr);
e6af00f1 BH	603	err = exofs_commit_chunk(page, 0, chunk_size);
e6af00f1 BH	604	fail:
09cbfeaf	605	put_page(page);
e6af00f1 BH	606	return err;
	607	}
	608
	609	int exofs_empty_dir(struct inode *inode)
	610	{
	611	struct page *page = NULL;
	612	unsigned long i, npages = dir_pages(inode);
	613
	614	for (i = 0; i < npages; i++) {
	615	char *kaddr;
	616	struct exofs_dir_entry *de;
	617	page = exofs_get_page(inode, i);
	618
	619	if (IS_ERR(page))
	620	continue;
	621
	622	kaddr = page_address(page);
	623	de = (struct exofs_dir_entry *)kaddr;
	624	kaddr += exofs_last_byte(inode, i) - EXOFS_DIR_REC_LEN(1);
	625
	626	while ((char *)de <= kaddr) {
	627	if (de->rec_len == 0) {
	628	EXOFS_ERR("ERROR: exofs_empty_dir: "
	629	"zero-length directory entry"
	630	"kaddr=%p, de=%p\n", kaddr, de);
	631	goto not_empty;
	632	}
	633	if (de->inode_no != 0) {
	634	/* check for . and .. */
	635	if (de->name[0] != '.')
	636	goto not_empty;
	637	if (de->name_len > 2)
	638	goto not_empty;
	639	if (de->name_len < 2) {
	640	if (le64_to_cpu(de->inode_no) !=
	641	inode->i_ino)
	642	goto not_empty;
	643	} else if (de->name[1] != '.')
	644	goto not_empty;
	645	}
	646	de = exofs_next_entry(de);
	647	}
	648	exofs_put_page(page);
	649	}
	650	return 1;
	651
	652	not_empty:
	653	exofs_put_page(page);
	654	return 0;
	655	}
	656
	657	const struct file_operations exofs_dir_operations = {
	658	.llseek = generic_file_llseek,
	659	.read = generic_read_dir,
c51da20c	660	.iterate_shared = exofs_readdir,
e6af00f1	661	};